Method for replacing contents of a data storage unit

Abstract

A data storage device includes a plurality of data storage units, a physical random number generator with a noise source based on a physical noise process, for generating a random number, and a replacer for selecting a data storage unit wherein data is to be stored, depending on the random number. Selecting, on the basis of genuine random numbers, data storage units and / or lines to be replaced in the cache.

Description

BACKGROUND OF THE INVENTION CROSS-REFERENCE TO RELATED APPLICATION [0001] This application is a continuation of copending International Application No. PCT / EP03 / 03800, filed Apr. 11, 2003, and is incorporated herein by reference in its entirety. [0002] 1. Field of the Invention [0003] The present invention relates to data storage devices, and in particular to secure operation of such data storage devices. [0004] 2. Description of Related Art [0005] Modern attacks on data processing means processing data relevant in terms of security, and / or attacks on the algorithm and secret keys processed therein are effected via so-called leak information. Leak information includes, for example, current consumption of the data processing unit, electromagnetic radiation during operation of the data processing means, etc. Conclusions about information relevant in terms of security may be drawn from a statistical analysis of the physical signals picked up. [0006] Here, the most common forms of attac...

AI Technical Summary

Benefits of technology

[0022] A further advantage of the present invention is the fact that no expensive software interventions and / or no additional current consumption is introduced by the random replacement strategy on the basis of actual physical random numbers.

[0023] A further advantage of the present invention is the fact that there is an option to combine the physical random number generator with a pseudo random number generator, which typically provides random numbers faster than a physical random number generator. By merely combining the physical random numbers and the pseudo random numbers, a good randomization of the current profile may be achieved, for example by infiltrating an actual random number after a certain number of pseudo random numbers, or by continuing to use the pseudo random numbers to select cache data storage units to be replaced, the feedback shift registers used for the pseudo random number generator obtaining random numbers as a seed or as seeds. Thus the replacement sequence will not always exhibit the same behavior at a restart of a CPU, but its behavior will in fact vary even though the pseudo random number generators which are fast and easy to implement, in terms of circuit technology, are nevertheless used. However, the pseudo random number generators are now placed into a new initial state (seed state) at selectable intervals by the physical noise generator.

[0024] A further advantage of the present invention is the fact that the use of a random number generator for cache replacement entails a significant degree of flexibility by providing scaleable security. For calculations highly relevant in terms of security, a cache line replacement may be performed on the basis of random numbers, whereas a switch-over from random number replacement to a fully deterministic replacement strategy may be effected by means of a mode control means and in accordance with certain criteria for data which are less relevant in terms of security but for which still a certain level of security is required.

Problems solved by technology

A disadvantage of this approach is the large-scale intervention in the respective software and the algorithms as well as the significant reduction in performance which results in most cases.

The benefit of this approach is doubtful, since in general it does not provide adequate protection against DPA or HO-DPA and additionally may lead to a significant increase in the current consumption of the data processing means.

Thus it may be made more difficult to find trigger points in current profiles and to perform a current profile analysis by means of resynchronization in the course of time.

Such a control typically comprises considerable complexity, since it must be ensured that the integrity of the data being processed and the integrity of the command flow are maintained despite the security measures.

With a high amount of cache associativity and large-scale memories, the random replacement strategy may achieve nearly the same performance as an LRU strategy, which is considerably more time-consuming and costly to implement.

However, this sequence is deterministic in that it is repeated in an exact manner whenever the predetermined seed has been fed into the LFSR.

In view of the above-described attacks on cryptographic data processing means, such pseudo random number generators and / or the random replacement strategy based on these pseudo random number generators have the significant disadvantage that the program flow will always remain deterministic.

Images